JPS6330665B2 - - Google Patents

Description

[Detailed Description of the Invention] The present invention relates to a character separation method, and in particular, even when a plurality of characters are written on a drawing etc. in such a way that they cannot be separated into individual characters even with a mask of a certain shape, these characters can be separated. This invention relates to a character separation method that allows each character to be separated into individual characters.

For example, when recognizing freely handwritten characters, the first step is to separate and extract each character, and then identify the extracted characters. When separating and extracting characters in this way, a mask M as shown in FIG. In this case, as shown in FIG. Since each character can be separated using a mask, it is possible to separate and extract these characters from each other.

However, when multiple characters are written close to each other as shown in Figure 1C, it is no longer possible to accurately separate these characters using the method of scanning a mask as shown in Figure 1B. It is possible and
Therefore, in such cases, it is difficult to accurately recognize characters.

Therefore, an object of the present invention is to provide a character separation method that can accurately separate each character even when a plurality of characters are written close to each other. Therefore, the character separation method according to the present invention includes an information holding means that holds image information of characters inputted from an input means, and an information holding means for holding image information of characters inputted from an input means, and scanning the image information in a first direction to obtain black and white change points. a first feature extracting means for detecting a first region that is a region between the two; a second feature extraction means for detecting a second region that is an area between points; and a detection means for detecting an intermediate point when scanning an area where the first area and the second area do not overlap in the first direction. Then, a first line is drawn in the second direction from the obtained intermediate point, and when the first line matches the character information, the first line is drawn in the second direction.
and a drawing line means for drawing a second line in the direction of , and the obtained first and second lines are used as separation lines that separate areas where individual character information exists.

Before explaining the present invention in detail below, the principle of the present invention will be explained based on FIGS. 2 to 9.

(a) First, memories m ₁ and m ₂ containing image data m ₀ obtained by reading a manuscript or drawing containing handwritten characters are prepared. Then, the memory _m1 is scanned in the x direction as shown in FIG. 2A. At this time, image data _m0 is obtained by setting the black dot where the characters are drawn as "1" and the blank area where no characters are drawn as "0". At this time,
As shown in Figure 2 (b), the change point is "0" →
Find "1" and the point where "1" changes to "0". For example, as shown in Figure 2 (b), the line l
When scanning above, x ₁ changes from "0" to "1", x ₂ changes from "1" to "0", x ₃ changes from "0" to "1", x ₄ Then "1" → "0"
Therefore, all of these x ₁ to x ₄ are changing points. In this case, the change point P ₁ (“1” →
0) and P ₂ (“0”→“1”), and all areas between the change points P ₁ and _{P 2} on the memory m ₂ are set to “1”.

(b) Next, the image data m ₀ on the memory m ₁ is transferred to the second
As shown in Figure A, scan in the y direction and similarly change the change point Q ₁ (“1” → “0” in the y direction) and
The pair Q ₂ (“0” → “1”) is detected, and the area on the memory m ₂ between Q ₁ and _{Q 2} is changed to “1” if “0” and “1” if “1”. ” and invert it to “0”. As a result, as shown in FIG. 3, the double line area becomes "0", the single line area becomes "1", and the data shown in FIG. 4 is written in the memory _m2 . At this time, since "1" is consecutive in the character part, it remains as is.

(c) Next, scan the memory m ₂ in which the data in the state shown in _FIG . ₂
Detect the pair (“1” → “0”). This can be detected by checking whether the area before the diagonal line has consecutive "0"s and whether there are "1"s consecutively more than the character width.

(d) Now, as shown in Figure 5, when the coordinates of R ₁ are (x ₁ , y ₁ ) and the coordinates of R ₂ are (x ₂ , y ₂ ), the center point M Find (xm, ym). In this case, xm = (x ₁ + x ₂ )/2, ym = y ₁
(y ₂ ). In this way, M ₁ to _{M 4} can be determined as shown in FIG. Then, a straight line is drawn in the y direction starting from these center points M ₁ to _{M 4} , and when it touches a character information point (“1” exists continuously), character information points are drawn again on both the left and right sides from that point. Draw horizontal lines until they touch. In this case, the fact that a horizontal line cannot be drawn from M ₄ is
It is clear from the figure.

(e) Next, as shown in FIG. 7, the memory m ₂ is scanned from below in the x direction. And as shown in Figure 8,
Similarly to (d) above, find the center points S ₁ to _{S 4} and draw a straight line from these in the y direction, and when this comes into contact with a character information point, draw a horizontal line in the same way. In this case S ₁ and
You cannot draw a horizontal line from S ₄ .

(f) At this time, as shown in Figure 8, the center point M ₁
If a vertical line from 2 intersects two horizontal lines p ₂ and l ₁ , draw a horizontal line lm between horizontal lines p ₂ and l ₁ . When the horizontal lines overlap in the x direction, the perpendicular lines v ₁ , v ₂ ,
v subtract ₃ .

(g) By drawing lines in this way, as shown in Figure 9, we can move the letters A, B, ......E to the center point M ₁ ,
M ₂ , M ₃ , S ₂ , S ₃ , horizontal line lm, p ₂ , l ₂ , p ₃ , l ₃ ,
Perpendicular lines v ₁ , v ₂ , v ₃ , etc. make it possible to distinguish them as a single character.

Next, one embodiment of the present invention will be described based on FIG. 10.

In the figure, 1 is the input section, 2 is the output memory, and 3 is the first
Image memory, 4 is a buffer, 5 is a first feature extraction section, 6 is a first address table, 7 is a first address generation section, 8 is a control section, 9 is a second address generation section, 10 is a buffer, 11 is a Second feature extraction unit, 1
2 is a second address table, 13 is a second image memory, 14 is an exclusive OR circuit, 15
is the third address generation part, 16 and 17 are buffers,
18 is a third feature extractor, 19 is a third address table, 20 is a boundary line extractor, 21 is a boundary line table, 22 is a fourth address generator, 23 and 24 are buffers, and 25 is a fourth feature extractor. It is.

The input unit 1 is an electrical signal generation unit that converts a handwritten manuscript or the like using, for example, a photoelectric conversion unit. Output memory 2
is a memory set for outputting single character information obtained by dividing information input from a handwritten manuscript or the like as shown in FIG.

The first image memory 3 is a memory in which image data input from the input section 1 is held. The buffer 4 is a working buffer memory for transmitting and holding the image data held in the first image memory 3, and is used to perform the processing as described in (a) above.

The first feature extraction unit 5 obtains a pair of change points P ₁ and P ₂ from the image data held in the buffer 4, as shown in (a) above, and sets the region of P ₁ to _{P 2} as “1”. The first address table 6 is a table in which the addresses of the areas of "1" between the change points _P1 and _P2 are held.

The first address generating section 7 generates an address for scanning the image data held in the buffer 4 in the x direction as shown in (a) above.

The control unit 8 processes the image data input to the first image memory 3 according to the steps (a) to (g) above, and performs various controls for creating individual character areas. For example, it controls the first address generation section 7 for scanning the buffer 4 or the first feature extraction section 5.

The second address generating section 9 generates an address for scanning the image data held in the buffer 10 in the y direction as shown in (b) above.
The buffer 10 is a memory to which the image data held in the first image memory 3 is sent and held, and is a working buffer memory for performing the processing as in (b) above.

The second feature extraction unit 11 extracts the change points Q ₁ and Q ₂ from the image data held in the buffer 10 as shown in (b) above.
, and reads out the Q ₁ and Q ₂ areas as "1".
2 is a table in which addresses of areas of "1" between the change points Q ₁ and Q ₂ are held.

The second image memory 13 is a memory in which the image data shown in FIG. 4 obtained as a result of the above (a) and (b) is set.

The third address generating section 15 generates an address for sequentially scanning the buffers 16 and 17 in the x direction from above the character. Batsuhua 16
is the fourth image set in the second image memory 13.
This is a buffer memory into which image data as shown in the figure is set, and the buffer 17 is a buffer memory into which the image data set in the first image memory 3 is set. These buffers 16 and 17 undergo the processing shown in (c) and (d) above to obtain center points M ₁ to M ₄ and vertical lines, horizontal lines, etc. generated therefrom as shown in FIG. Then, the character is sequentially scanned in the x direction starting from the top.

The third feature extraction unit 18 extracts the above (c) and
This process performs the process (d). That is, the buffer 1 in which the image data shown in FIG.
6 and the buffer 17 in which the image data transmitted from the first image memory 3 is set are sequentially scanned in the x direction from above the character based on the address information generated from the third address generation section 15. The output data of is sequentially transmitted to the three feature extraction sections 18. Then, based on the data output from the buffer 16, the above-mentioned inflection point within the shaded area in Fig.
Detect R ₁ and R ₂ . Then find the center point and let it be M ₂ . In this way, convert the buffer 16 to x
Center points M ₁ , M ₃ , and M ₄ are obtained by scanning sequentially in the direction as described in (c) above. At this time, character information is transmitted from the buffer 17, so unlike the example described in (c) above, character positions are identified using this information. After obtaining the center points M ₁ to _{M 4} in this way, the third feature extraction unit 18 uses its drawing line circuit 18-0 to draw a straight line downward in the y direction from these center points M ₁ to _{M 4} . Pull. When this contacts a character (of course, the character position is obtained from the character information transmitted from the buffer 17), a straight line is drawn from the contact point in the left and right x directions. Draw each of these lines until they touch the letters. Therefore, only a vertical line in the y direction is drawn at the center point _M4 .

The third address table 19 includes the above (c), (d),
This is a table in which the center points M ₁ to _{M 4} and S ₁ to _{S 4} obtained in (e), the coordinates of each straight line in contact with the characters, the coordinates of the intersection of each straight line, etc. are entered.

Based on the data transmitted from the third address table 19, the boundary line extraction unit 20 extracts the
By performing the processing in (f) and (g), boundaries between characters as shown in FIG. 9 are created. The boundary line table 21 is a table in which data of the boundary positions between each character created by the boundary line extraction section 20 is entered.

The fourth address generating section 22 generates addresses for sequentially scanning the buffers 22 and 23 in the x direction starting from the bottom of the character, as explained in (e) above. The buffer 23 is a buffer memory into which the image data as shown in FIG. 4 set in the second image memory 13 is set, and the buffer 24 is into which the image data set in the first image memory 3 is set. buffer memory. And these batshua 2
3 and 24 are the 8th after the process shown in (e) above is performed.
In order to obtain the center points S ₁ to S ₄ and the vertical lines, horizontal lines, etc. generated therefrom as shown in the figure, the characters are sequentially scanned in the x direction from the lower position of the character.

The fourth feature extractor 25 performs the process (e) above based on the image data set in the buffers 23 and 24. That is, the buffer 23 in which the image data shown in FIG. ,
The character is sequentially scanned in the x direction from below, and the output data is sequentially transmitted to the four feature extraction sections 18. This fourth feature extraction unit 18 obtains the center points S ₁ to _{S 4} based on the data transmitted from the buffer 23, and uses the drawing line circuit 25-0 to extract these center points.
Draw a straight line upward in the y direction from S ₁ to _{S 4} . When this comes into contact with a character (of course, this character position is obtained from the character information transmitted from the buffer 24),
Next, draw a straight line in the x direction to the left and right from the contact point. Each of these straight lines is drawn until it touches _the characters, so as shown in _{FIG .} A straight line p ₂ is drawn for the center point S ₃ , and a straight line p ₃ is drawn for the center point S 3 .

The operation shown in FIG. 10 will be briefly explained below.

(1) First, image information such as a handwritten manuscript is converted into an electrical signal by the input section 1 to become image data of "1" and "0", and is set in the first image memory 3. This image data is then sent to the buffers 4 and 10 and set there as well.

(2) After the image data is set in the buffers 4 and 10 in this way, the control section 8
For the address generation section 7, add a buffer of 10 x
The buffer 10 is controlled to generate an address so as to scan in the y-direction. As a result, the first feature extraction unit 5
detects the pair of changing points P ₁ and P ₂ shown in (a) above and sets the area of this pair of changing points P ₁ to _{P 2} as “1”,
This address area set to "1" is set in the first address table 6. On the other hand, the second feature extraction unit 11 detects the pair of changing points Q ₁ and Q ₂ shown in (b) above, and sets the area of this pair of changing points P ₁ to P 2 _to “1”.
and set this address area as "1" to the second address area.
Set in address table 12.

(3) In this way, the change points P ₁ ~ P ₂ and Q ₁ ~ Q ₂
After setting the area between them to "1", the exclusive OR circuit 14 sets "1" in the first address table 6 and the second address table 12.
Taking the exclusive OR of the area, image data as shown in Fig. 4 is obtained.
This is set in the second image memory 13.

(4) The image data shown in FIG. 4 obtained in this way is transferred from the second image memory 13 to the buffer 16.
and 23, and the original image data set in the first image memory 3 is transferred to the buffers 17 and 24.

(5) Then, the control unit 8 controls the third address generation unit 15
On the other hand, an address is generated for normal scanning in which the buffers 16 and 17 are sequentially scanned from above the character in the x direction. Based on the data thus output, the third feature extraction unit 18 extracts the change points R ₁ and R ₂ as shown in (c) above.
As shown in Fig. 6, the center point M ₁
Find ~M ₄ and draw a vertical line in the y direction from this,
When this touches a character, draw a straight line horizontally in the x direction from that point. Then draw this horizontal line until it touches the letters. At this time, character information can be obtained from the buffer 17.

(6) At the same time, the control unit 8 instructs the fourth address generation unit 22 to scan the buffers 23 and 24 from below in order to sequentially scan in the x direction from below the character, as shown in FIG. Generate an address. Based on the data outputted from this, the fourth feature extracting section 25 finds a similar change point and thereby obtains the center point S ₁ as shown in FIG.
~Get _S4 . Then, straight lines are drawn upward in the y direction from these center points S ₁ to S ₄ . And when this comes into contact with the character, this time the x to the left and right of that contact point
Draw a straight line in the direction. Draw this horizontal straight line until it touches the letters. The character information at this time can be obtained from the buffer 24.

_{( 7 )} In this way _, as shown in FIG. ₁
Data such as ~l ₃ , p ₂ , p ₃ etc. are sent to the third address table 19.

(8) Based on the data sent from this third address table 19, the boundary line extraction unit 20
For lines that intersect the horizontal lines p ₂ and l ₁ , such as the line in the y direction from the center point M ₁ , a horizontal line lm is drawn in the center of them, and the lines l ₂ , p ₂ ; ₂ , p ₃ and l ₃ , p ₃ mutually x
For areas with overlapping directions, vertical lines v ₁ , v ₂ ,
v subtract ₃ . In this manner, the characters A to E are divided into individual character areas as shown in FIG.

(9) The coordinate information divided as above is set in the boundary line table 21, and based on this, the first
The image data set in the image memory is extracted one character at a time and sent to the output memory 2, from which it is sent to, for example, a character recognition device, where the characters are identified.

As described above, according to the present invention, even when a plurality of characters are written close to each other, it is possible to separate them into individual characters. Therefore, even handwritten characters that are not written within a predetermined frame on the manuscript paper can be separated individually. Therefore, since even such handwritten characters can be separated and extracted character by character, great effects can be exerted when, for example, character recognition of handwritten characters is performed.

[Brief explanation of the drawing]

Fig. 1 is an explanatory diagram of a mask for character reading and cases in which character separation is possible and not possible using the mask, Fig. 2 is an explanatory diagram of the scanning state, and Figs. 3 to 9 are diagrams illustrating the operation of the present invention. The state explanatory diagram, FIG. 10, is a configuration diagram of an embodiment of the present invention. In the figure, 1 is the input section, 2 is the output memory, and 3 is the first
Image memory, 4 is a buffer, 5 is a first feature extraction section, 6 is a first address table, 7 is a first address generation section, 8 is a control section, 9 is a second address generation section, 10 is a buffer, 11 is a Second feature extraction unit, 1
2 is a second address table, 13 is a second image memory, 14 is an exclusive OR circuit, 15
is the third address generation part, 16 and 17 are buffers,
18 is a third feature extractor, 19 is a third address table, 20 is a boundary line extractor, 21 is a boundary line table, 22 is a fourth address generator, 23 and 24 are buffers, and 25 is a fourth feature extractor. are shown respectively.

Claims (1)

[Claims] 1. Information holding means for holding image information of characters inputted from an input means, and a first area that scans the image information in a first direction to obtain points of black and white change, and is an area between the points of change. a first feature extraction means for detecting a black and white change point by scanning the image information in a second direction substantially orthogonal to the first direction, and a second feature extracting means for detecting a black and white change point of the image information; a second feature extraction means for detecting, a detection means for detecting an intermediate point when scanning an area where the first area and the second area do not overlap in the first direction, and a second feature extracting means for detecting an intermediate point from the obtained intermediate point. drawing means for drawing a first line in the second direction and drawing a second line in the first direction when the first line matches character information; A character separation method characterized in that the first and second lines are used as separation lines that separate areas where individual character information exists.